Measurements, Data Analysis and Control Methods in Wastewater Treatment Plants–State of the Art and Future Trends

Abstract

This paper is a summary of a committee working for the Swedish Water and Wastewater Works Association (VAV). The purpose of the report is to present the possibilities today to measure, present and analyze data and control treatment plants. The typical audience is the operator, the process engineer, or the consulting engineer. The methods presented are all known from different disciplines, but are here presented in a form that connects the methods to wastewater treatment operation. Unlike any manual of practice the report is not a concensus report of current practice-Rather it is an attempt to show the potential of modern methods for data analysis and control. This will help the potential equipment or computer buyer to specify relevant demands for the system. The fact that any wastewater treatment plant is highly dynamic has to be reflected both in measurements and in control. The report discusses relevant sampling times for different measurements, both from the inherent dynamics and from the variability of the disturbances. Current design practice is almost always based on steady state analysis, and disturbances are too often controlled by larger tank volumes rather than relevant control actions. In order to obtain relevant data analysis the purpose of the measurement has to be clearly stated. Interesting and relevant measurement variables are listed. Moreover, a short survey of existing instrumentation and its status is presented. The transfer of data from the primary sensor to the computer has to be carefully designed. Once the data is in the computer, the data structure must be specified. The different compromizes between storage capacity, data formats and other relevant information are discussed. Simple measurement handling is described before statistical analysis is discussed. Numerous examples demonstrate the results. Some methods for parameter estimation and model building from measurement data are discussed, particularly with the purpose to make the methods available for on-line use. It is shown how estimated models can be used for the operation of plants. Different control methods are discussed. The basic kind of local control to keep the plant running is first mentioned, but more emphasis is laid on plant quality control, like dissolved oxygen, return sludge and waste sludge control. Dynamic models offer interesting possibilities for plant simulation, and simulators are being developed, that can support the operator with further predictive information. Some future possibilities of knowledge-based systems for process diagnosis are further discussed. They offer new possibilities to use natural language for systematic error analysis and diagnostic searches.